automobile fuel economy standards: impacts, efﬁciency, and ...nature.berkeley.edu › ~sallee ›...

Symposium:Transportation and the Environment

Automobile Fuel Economy Standards:Impacts, Efficiency, and Alternatives

Soren T. Anderson,* Ian W. H. Parry,y James M. Sallee,z andCarolyn Fischery

Introduction

A number of countries regulate the fuel economy of new light-duty vehicles, while othersregulate carbon dioxide (CO2) emission rates per mile, which is almost equivalent. Thereare two main economic rationales for these regulations.The first rationale is that fuel economy standards require automakers to design more

efficient vehicles or to shift sales toward more efficient models. This lowers CO2 emissionsand reduces dependence on oil markets that are subject to economic and political uncertainty.However, these regulations are aimed only at improving the fuel economy of new automo-biles; they do not encourage other forms of conservation or affect other sectors. In contrast,direct taxes on oil or on carbon raise fuel prices for everyone. Thus, such taxes promote fueleconomy in new automobiles, discourage driving by owners of new and used vehicles alike,and reduce emissions and oil use beyond the automobile sector. This is important becauseother sectors play a key role in both CO2 emissions and oil consumption. In fact, automobilesaccount for ‘‘only’’ about 20 percent of CO2 emissions and 45 percent of oil use, both in theUnited States and worldwide (EIA 2009; IEA 2009).The second rationale is that fuel economy standards may address a market failure that arises

because consumers misperceive the benefits of improved energy efficiency. This rationale is con-troversial, however, because information dissemination programs may be more efficient thanstandards in dealing with this issue. Moreover, while a number of empirical studies find thatconsumers, in a variety of settings, appear not to pursue cost-effective opportunities to improveenergy efficiency to the extent that they should, this finding is not universal (see, e.g., Helfand

*Department of Economics and Department of Agricultural, Food, and Michigan State University;e-mail: [email protected] for the Future; e-mail: [email protected]; [email protected] Harris School, University of Chicago; e-mail: [email protected].

We are grateful to Gloria Helfand, Mark Jacobsen, Suzy Leonard, Joshua Linn, Kenneth Small, and a refereefor very helpful comments on an earlier draft.

Review of Environmental Economics and Policy, volume 5, issue 1, winter 2011, pp. 89–108doi:10.1093/reep/req021Advance Access publication on May 24, 2011! The Author 2011. Published by Oxford University Press on behalf of the Association of Environmental andResource Economists. All rights reserved. For permissions, please email: [email protected]

89

at University of C

hicago Library on August 4, 2011reep.oxfordjournals.org

Dow

nloaded from

http://reep.oxfordjournals.org/

and Wolverton 2009; Greene 2010). In addition, even when consumers neglect apparently cost-effective opportunities, this behavior may reflect unmeasured costs rather than a market failure.Whether or not there is a strong efficiency rationale for fuel economy standards, the fact

that they are widely implemented suggests that they may be more politically tractable thanother types of policy instruments. This fact raises a number of important policy questions.The purpose of this article, which is part of a symposium on Transportation and the Envi-ronment,1 is to assess the impacts and efficiency of automobile fuel economy standards andcompare them to alternative policy instruments such as fuel taxes and ‘‘feebate’’ systems,which impose fees on inefficient vehicles and provide rebates for efficient vehicles accordingto their fuel consumption per mile.We begin by discussing the structure of fuel economy standards in practice; their influence

on fleet composition, driving, and other behaviors; and their past and likely future effective-ness. We then review engineering and market-based approaches to measuring the costs offuel economy programs. Next we compare the welfare effects of fuel economy standardsand fuel taxes and assess whether these two policies can complement each other. Given thathigher fuel taxes are widely believed to be politically infeasible in the United States, wenext evaluate the case for replacing fuel economy regulations with feebate systems. The finalsection summarizes our findings and offers some conclusions.Because U.S. studies dominate the literature and most other countries have only recently

begun to implement similar policies, our discussion focuses largely on fuel economy regu-lations in the United States. However, whenever possible, we draw comparisons with, andhighlight broad implications for, other countries.

Fuel Economy Standards in Practice

This section describes past and present fuel economy standards in the United States and othercountries, examines how these regulations operate, and assesses their impacts.

Fuel Economy Standards in the United States

The Corporate Average Fuel Economy (CAFE) program was introduced in the United Statesin 1975 with the primary goal of reducing the country’s dependence on foreign oil. Auto-makers were initially required to meet a sales-weighted average of 18 miles per gallon (mpg)for their car fleets, which increased steadily to 27.5 mpg by 1985. A lower standard was estab-lished for light trucks (pickups, minivans, and sport utility vehicles [SUVs]), which rose from16 mpg in 1980 to 22.5 mpg in 2008. The rationale for this lower standard was that truckswere used primarily by businesses and farmers as part of their operations. However, this isobviously no longer the case.As a result of the passage of the Energy Independence and Security Act (EISA) of 2007 and

administrative action by the Obama Administration in 2009, fuel economy standards will beaggressively tightened between 2011 and 2016. In fact, there are now two separate regulations,one governing fuel economy, issued by the National Traffic Highway Safety Administration

1The other articles include Proost and Van Dender (2011), who review projections of road transport demandand assess long-term policy issues and options, and Anas and Lindsey (2011), who examine the use of roadpricing as a tool for reducing congestion and other externalities related to urban road transportation.

90 Anderson et al.

at University of C


Dow

nloaded from


(NHTSA), and the other regulating CO2 emissions per mile, issued by the Environmental Pro-tection Agency (EPA). The standards are essentially equivalent, except that the slightly tighterEPA requirement allows automakers to earn compliance credits by modifying air conditionerrefrigerants to reduce greenhouse gases. Without the air conditioner credit, EPA’s standardwould yield a combined average fuel economy for new passenger vehicles of 35.5 mpg(250 g CO2 per mile) in 2016 (assuming no change in the shares of car versus truck sales).This compares to a combined standard of about 25 mpg in 2008. Automakers currentlypay a fine of $55 per vehicle for every 1 mpg that their fleet average falls short of the relevantstandard. Presumably this fine will be increased to enforce the stricter future standards.The structure of the CAFE program is also being radically reformed, though some details are

still being finalized. Each vehicle will face a separate fuel economy target based on its size orfootprint.2 There are separate mathematical formulas for cars and light trucks, which convertthe footprints of individual models into fuel economy targets. These targets are then used tocalculate sales-weighted standards for each automaker’s car and light-truck fleet. Automakersspecializing in smaller vehicles will face more stringent standards, reducing incentives for auto-makers to downsize vehicles to comply with regulation. Other pending changes to the CAFEprogram include expanding opportunities for automakers to bank and borrow fuel economycredits, allowing automakers to transfer credits between their car and light-truck fleets, and allow-ing credit trading across firms. The implications of such flexibility provisions are discussed below.One looming issue (not addressed in detail here) is how fuel economy standards might be

adapted for electric and alternative-fuel vehicles, which promise to play an increasinglyimportant role in energy strategies throughout the world.3 Accounting properly for vehiclesthatdrawelectricity fromthegrid, includingplug-inhybrids andelectric vehicles, isnoeasy tasksince it requires attributing some levelof emissions toelectricity.These emissionsdependon thefuel used to generate extra electricity,whichvaries greatly by locationand timeofday.As regardsalternative fuels, theCAFE standards include a loophole that treats flexible fuel vehicles capableof burning either gasoline or ethanol as though they run on ethanol 50 percent of the time eventhough, in practice, these vehicles rarely use ethanol (Anderson and Sallee 2010). Although thisloophole is scheduled to phase out by 2020, regulators still need to establish awayofmeasuringthe true gasoline consumption or CO2 emissions of vehicles that use alternative fuels.

Fuel Economy Standards in Other Countries

Figure 1 summarizes the fuel economy standards in the United States and other countries thathave similar programs.4 In the European Union (EU), in response to regulations that set anultimate target of 130 g CO2 per kilometer, average new vehicle fuel economy is set to reach 45mpg in 2012 and to continue rising thereafter. Tighter standards are easier to meet in Europebecause high fuel taxes and the predominance of small cars and more fuel-efficient diesel

2A vehicle’s footprint is its track width (distance between the centers of the left and right wheels) multipliedby its wheel base (distance between the front and rear axles).3In part, this growth in the importance of electric and alternative-fuel vehicles will be policy driven. Forexample, the EISA strengthened the requirements for blending biofuels with gasoline. Low carbon fuel stand-ards (recently introduced in California) will also encourage biofuels. There could be interesting interactionsbetween these sorts of policies and fuel economy regulations that have yet to be explored in the literature.4For more details on these programs, see An and Sauer (2004), Elmer and Fischer (2010), and IEA (2009).

Automobile Fuel Economy Standards 91

at University of C


Dow

nloaded from


engines imply a higher baseline fuel economy. Unlike the United States, which has separatestandards for cars and trucks, the EU has one set of regulations for the entire light-duty fleet,but a so-called ‘‘limit value curve’’ allows heavier cars to have higher emissions than lightercars while preserving the overall fleet average. As part of the phase-in of the new regulations,the EU penalties for noncompliance are applied on a sliding scale through 2018, with lowpenalties of V5 for the first g/km in excess of the standard, which rise to V95 for the fourthg/km in excess and beyond.China sets maximum fuel consumption standards for each vehicle, based on weight rather

than average fleet-wide standards. Japan sets different fuel efficiency standards for diesel andgasoline vehicles, further differentiated by weight class. The future targets in Japan are basedon the current ‘‘top performer’’ in each weight class (excluding niche products). Although thetargets in Japan are mandatory, compliance seems to rely heavily on social pressures, as mon-etary penalties are low. Canada has modeled its fuel economy standards on those in theUnited States, but compliance is voluntary.As in the United States, fuel economy programs in other countries are becoming more flexible.

Examples include themovement towardweight-based standards in the EU, Japan, andChina. TheEU also allows pooling of targets across manufacturers to comply jointly with the standard. Japanallows manufacturers to accumulate credits in one weight category for use in another.

How Fuel Economy Standards Operate

Predicting the effects of fuel economy programs is inherently difficult, given the complexity ofnew vehicle markets, in which firms exercise market power, provide several differentiated

Figure 1 Fuel economy standards for new passenger vehicles by countrySource: ICCT (2009).Notes: Dashed lines indicate proposed fuel economy targets not yet enacted. For Canada, the programincludes in-use vehicles.

92 Anderson et al.

at University of C


Dow

nloaded from


products, and bundle many attributes into a single good. Firmsmay not behave as price takersbecause there is considerable market concentration, opening the door for strategic pricingthat accounts for the behavior of competitors.5 Typically, firms offer a variety of models to thesame consumer base, which complicates strategic pricing—a firm that lowers the price on oneof its models will not only siphon customers from its competitors but also lower demand forits other models. Besides price, automakers choose many physical attributes to bundle intoa particular model. An equilibrium in this market, which is usually modeled as a Nash out-come in which firms cannot profitably deviate by changing prices or vehicle attributes giventhe choices made by other firms (e.g., Austin and Dinan 2005; Bento et al. 2009; Jacobsen2010a), requires firms to choose attributes, fuel economy, and prices for multiple vehicles.How do fuel economy standards influence these choices? When fuel economy standards are

binding, firms must deviate from their profit-maximizing plan and raise their fleet averagempg to meet the minimum imposed by the standard. To do this, firms can alter vehicle char-acteristics—either by incorporating costly fuel-saving technologies or by compromising othervehicle attributes, such as size and horsepower—or change relative prices to increase the salesshares of more efficient models. Vehicle redesign is likely responsible for most improvementsin fuel economy to date since studies find that compliance costs are substantially lower in thelong run when automakers can redesign vehicles to be more efficient (e.g., Jacobsen 2010a;Klier and Linn 2010; Whitefoot, Fowlie, and Skerlos 2010).Fuel economy regulations place an implicit tax on inefficient vehicles and changes in vehicle

attributes that reduce fuel economy, while subsidizing efficient vehicles and changes thatimprove fuel economy. With separate standards for cars and light trucks, these taxes andsubsidies operate separately within these two fleets, meaning that large cars are taxed whilepotentially less efficient small trucks are subsidized. This creates a perverse incentive toredesign large cars as trucks (e.g., the Chevy HHR and Chrysler PT Cruiser).6 Moreover,in the absence of provisions that allow credit trading between firms, there are no implicittaxes or subsidies on firms like Honda and Toyota, which perennially exceed U.S. standards.This means that even if tighter standards cause low mpg manufacturers (e.g., Ford andGeneral Motors) to improve fuel economy, their high mpg competitors (e.g., Honda andToyota) may expand their offerings to capture the consumers who desire large vehicles.Whitefoot, Fowlie, and Skerlos (2010) estimate that this ‘‘leakage’’ is substantial.Tighter fuel economy standards affect only new vehicles. Thus, when the standard

increases, overall fuel economy improves gradually for about fifteen years as new vehiclesreplace the old. By lowering fuel costs per mile, fuel economy standards also encourage moredriving. Recent evidence for the United States suggests that this ‘‘rebound effect’’ is fairlymodest, however, offsetting just 10 percent of the fuel savings resulting from higher fuel econ-omy (e.g., Small and Van Dender 2007). Both the rebound effect and the gradual turnovercause fuel taxation to be more efficient than fuel economy standards in conventionaleconomic models (e.g., Austin and Dinan 2005).

5The seven largest firms in the U.S. market in 2009 (General Motors, Toyota, Ford, Honda, Chrysler, Nissan,and Hyundai) accounted for 87.5 percent of sales (Automotive News 2010).6This increases an automaker’s average fuel economy for bothfleets since a redesigned vehiclewouldhave fallenshort of the car standard but exceeds the truck standard. Footprint-based standards will partially correct thisperverse incentive, however, because small trucks will have fuel economy targets above the truck average.


at University of C


Dow

nloaded from


Impacts of Fuel Economy Standards

The United States has the longest history with fuel economy regulation, and thus our dis-cussion focuses on U.S. experience. Given the complexities described above and the difficultyof distinguishing the effects of regulation from changing fuel prices and preferences for ve-hicle size, it is not possible to make precise statements about the historical impacts of CAFE.However, a few stylized facts are clear.First, for most of its existence, CAFE has been binding. As a result, actual fuel economy for

new vehicles closely tracked the standard between 1978 and 2000 (see Figure 2a), especiallyduring the 1990s when the CAFE standards were unchanged.7 Only during the run up in fuelprices in the 2000s were standards potentially nonbinding.Second, while engine efficiency has improved during the last two decades, automakers have

sacrificed potential improvements in fuel economy to make bigger, more powerful cars, pre-sumably in response to perceived consumer preferences. Figure 2b shows the evolution of twokey performance measures: (1) horsepower divided by weight; and (2) time to accelerate to 60miles per hour. As fuel economy increased rapidly during the initial phase-in of CAFE stand-ards, performance was flat. Then, for about twenty years, as CAFE standards and fuel economyboth stabilized, automakers steadily improved performance. During that time, there were sig-nificant advances in fuel efficiency (in the sense of energy harnessed per gallon of fuel combus-tion), but the gains were allocated away from fuel economy (in the amount of fuel necessary totravel some distance), enabling bigger and faster vehicles to meet the CAFE requirements.Third, the share of light trucks in new vehicle sales increased from 3 percent in 1978 to

about 50 percent in 2003, causing overall fuel economy for new vehicles to fall slightly duringthe 1990s (Figure 2c). A large part of this change was due to the introduction of minivans,which replaced stationwagons, and then the rise of SUVs as family cars. Changes in consumerpreferences are at least part of the explanation for the growth in light-truck sales, which beganbefore implementation of CAFE (Davis 1999). However, the CAFE standards likely contrib-uted to this transition by creating incentives to design car-like vehicles to qualify as trucks(Sallee 2010; Sallee and Slemrod 2010).The impact of fuel economy standards on road safety is less clear (e.g., Jacobsen 2010b). To

the extent that it encouraged light-truck sales, CAFE likely reduced traffic fatalities for thosetruck occupants who would have otherwise bought cars but increased fatalities for everyoneelse (e.g., Gayer 2004; Li 2010). White (2004) finds that for each fatal crash that occupants oflarge vehicles avoid, more than four additional fatal crashes occur that involve others. On theother hand, for a given fleet mix, fuel economy regulations probably lead to smaller andlighter vehicles overall, which are less safe in single-vehicle collisions but can reduce injuryrisks to others in multi-vehicle collisions. Moreover, from an economic efficiency perspective,what matters are the overall external costs of traffic accidents, rather than highway fatalities.External costs exclude the risk of injuring oneself in an accident but include injuries to othersand third-party property damage, medical costs, and productivity losses. However, based onthe available literature, it is difficult to draw definitive conclusions about the direction,

7This is consistent with the claims of the ‘‘big three’’ U.S. automakers, who have argued that CAFE is bindingand costly for them. Broadly, the Asian automakers have historically been well above the standard, whileEuropean luxury-car makers have frequently paid noncompliance penalties.

94 Anderson et al.

at University of C


Dow

nloaded from


let alone the magnitude, of the link between external accident costs and fuel economyregulations.

Likely Future Impacts

Looking to the future, according to NHTSA (2010), automakers can meet future CAFErequirements withmoderately costly technologies (e.g., cylinder deactivation, turbo charging,engine downsizing, conversion to dual-clutch transmissions, and start–stop engine

Figure 2 Trends in U.S. fuel economy and other vehicle characteristicsSource: EPA (2009), NHTSA (2009).


at University of C


Dow

nloaded from


technology) and other modifications (e.g., weight reductions), without causing a deteriora-tion in power, acceleration, or other attributes.All else equal, raising average new vehicle fuel economy from 25 to 35 mpg would reduce

long-run fuel consumption by about 30 percent. Disentangling the actual future effect oftighter standards is difficult, however, given that future fuel use would likely fall anyway(reducing the effect of the standards) with rising fuel prices, biofuel mandates, and a shiftin sales back toward cars. Small (2010) considers a continued tightening of fuel economyregulations beyond 2016 that ultimately raises average new vehicle fuel economy to 46mpg by 2030. In this scenario, by 2030 gasoline use falls 23 percent below 2010 levels, despitea projected 33-percent increase in vehicle miles traveled over the period. However, at this point,Small (2010) finds that manufacturers are likely to run out of viable technologies and mayprefer to pay fines in lieu of maintaining full compliance with CAFE. This underscores thetechnological constraints on the ability of standards to continue to increase fuel economy.

Assessing the Costs of Fuel Economy Standards

There are two broad methodological approaches to assessing the costs of fuel economy stand-ards (excluding their impacts on externalities). The engineering approach envisions auto-makers adding fuel-saving technologies to existing models, taking as given other vehicleattributes, such as power, and the size and composition of the new vehicle fleet. In contrast,the market-modeling approach accounts for broader behavioral responses, including alteringthe sales mix of new vehicles and improving fuel economy at the expense of other attributes.

The Engineering Approach and Its Limitations

Under this approach, upfront costs and per-mile fuel savings are assessed for a wide range oftechnologies (see, e.g., NRC 2002; Creyts et al. 2007; NHTSA 2010). Automakers are assumedto progressively adopt the most cost-effective technologies to meet fuel economy require-ments. Estimates of lifetime fuel savings are then subtracted from the technology adoptioncosts.8 NHTSA (2010) has estimated that its new standard would add about $900 in incre-mental technology costs for the average new vehicle in 2016 but generate about $3,200 in fuelsavings and other private benefits (e.g., reduced refueling time), implying a negative netprivate cost of about $2,300 per vehicle. Although NHTSA (2010) uses a relatively low socialdiscount rate of 3 percent, private costs are still negative even under a 7-percent rate. Note thatwe have defined benefits here to exclude externalities.These estimates of negative costs beg the question of why automakers have not already

incorporated these seemingly profitable technologies. One possibility, put forward byNHTSA, is that consumers undervalue fuel economy benefits because of myopia, unreason-ably short planning horizons, imperfect information, bounded rationality, limited impor-tance of fuel costs in vehicle purchases, or simply an inability to calculate the financialbenefits of energy efficiency properly.

8Pretax fuel prices should be used here, as savings in tax payments to motorists are offset by a revenue loss tothe government. To the extent that the rebound effect is included, it plays a relatively minor role (leaving asideexternalities), as smaller savings in lifetime fuel costs are approximately offset by increased driving benefits.

96 Anderson et al.

at University of C


Dow

nloaded from


Skeptics of this ‘‘misperceptions’’ market failure see no reason why consumers should besystematically misinformed, not least because EPA fuel economy estimates are required to beprominently displayed on new vehicles. According to these skeptics, various ‘‘unmeasured’’costs (i.e., costs that are unobserved or ignored in engineering studies) account for any dif-ference between the present discounted value of fuel savings and the cost of fuel-saving tech-nologies. Unmeasured costs might include costs associated with actually implementing newtechnologies, such as marketing, maintenance, and retraining mechanics. They also mightinclude the opportunity costs of using emerging engine technologies to enhance fuel econ-omy at the expense of competing vehicle attributes (such as horsepower or additional energy-using devices) that consumers value more highly (e.g., Austin and Dinan 2005; Fischer,Harrington, and Parry 2007; Knittel 2009). A further unmeasured cost may arise if risk-averseconsumers discount fuel economy benefits because they are uncertain about future fuel pricesor actual fuel savings.9 If such unmeasured costs were included in the analysis, the argumentgoes, then the apparent net private benefit of tighter CAFE standards would disappear.Research on the misperceptions market failure hypothesis remains inconclusive. Lab

experiments suggest that consumers are confused by the fact that cost savings are linearin gallons per mile—not miles per gallon, which is the standard way of reporting vehicleefficiency in the United States (e.g., Larrick and Soll 2008)—while qualitative surveys showthat many consumers know little about the mpg of their current vehicles, their future drivingpattern, and how to discount fuel savings (e.g., Turrentine and Kurani 2007). While thesefindings suggest that consumers may inaccurately assess fuel economy benefits, the directionof any bias is unclear. Tests of market behavior provide an evenmurkier picture. For example,Dreyfus and Viscusi (1995) estimate implicit discount rates for new vehicle fuel economy ofup to about 20 percent, which is broadly consistent with estimated implicit discount rates forother energy durables (e.g., Hausman 1979; Dubin andMcFadden 1984). However, the rate ofa private automobile loan was around 10–15 percent during their sample period, suggestingthat high implicit discount rates may largely reflect borrowing costs. Given that credit marketsfor automobile loans are extensive and competitive, these high rates may simply reflect defaultrisks, rather than any market failure. Another branch of the literature uses fuel price variationto test whether vehicle prices adjust by as much as predicted when consumers fully value fueleconomy, but again the findings are mixed (e.g., Kahn 1986; Kilian and Sims 2006; Allcott andWozny 2009; Helfand and Wolverton 2009; Sallee, West, and Fan 2009; Greene 2010).

Market-Modeling Approach

Market-modeling studies may address some of the limitations of engineering approaches,which ignore the impact of fuel economy standards on fleet size and composition and onother vehicle attributes, such as weight and power.Market-modeling studies simulate the effects of CAFE regulations on gasoline consump-

tion, automaker profits, and consumer welfare using explicit models of the new vehiclemarket. In these studies, vehicle production costs typically depend on fuel economy andare based on widely used technology cost assessments such as NRC (2002) or possibly

9Fuel economy ratings are inexact—a factor highlighted by the EPA’s recent overhaul of the rating system;individual fuel economy performance varies significantly depending on driving behavior (Sallee 2010).


at University of C


Dow

nloaded from


econometric estimates that are based on automaker behavior. Consumer demand functionsfor new vehicles are based on either detailed econometric models, in which household choicesdepend on prices and a range of vehicle attributes (e.g., Goldberg 1998; Gramlich 2008;Jacobsen 2010a; Klier and Linn 2010), or assumed own- and cross-price vehicle demandelasticities (e.g., Kleit 2004; Austin and Dinan 2005; Fischer, Harrington, and Parry 2007;Goulder, Jacobsen, and van Benthem 2009; Small 2010). Fuel economy improvements affectvehicle demand via the pass-through of technology adoption costs into prices and throughconsumer preferences for fuel economy and other attributes. For tractability, most studiesassume that automakers compete on prices in a Nash–Bertrand setting and that profits aremaximized on a year-by-year basis. Most studies also model the rebound effect.10 Becausemarket-modeling studies differ considerably in their assumptions and methodologies and itis difficult to judge which models are the most reliable, we focus here on qualitative ratherthan quantitative findings concerning the costs of fuel economy standards.First, in contrast to NHTSA (2010), virtually all recent market-modeling studies find that

CAFE standards impose nonnegligible costs on automakers and consumers. However, thesestudies rule out the consumer misperceptions hypothesis by assuming that the private marketfor fuel economy operates efficiently in the absence of CAFE standards.11

Second,market-modeling studies find that the short-run cost estimates for a small increase inthe CAFE standard typically exceed long-run cost estimates by a factor of 2–3 (e.g., Jacobsen2010a; Klier and Linn 2010; Whitefoot, Fowlie, and Skerlos 2010). This reflects the fact that inthe longer run, automakers have greater scope for altering vehicle characteristics tomeet a givenstandard, while in the very short run their only compliance option is to alter the sales mix.Third, market-modeling studies indicate that gasoline taxes are a far more cost-effective

policy than CAFE standards because they exploit more options for reducing gasoline use.Austin and Dinan (2005) and Jacobsen (2010a) estimate that for a given long-run reductionin fuel consumption, CAFE standards are about two to three times more costly than a gasolinetax. Jacobsen (2010a) finds that total welfare costs average about $2 per gallon of fuel savedfor a 1 mpg increase in the CAFE standard, while a gasoline tax that saves the same amount offuel imposes welfare costs of about $0.80 per gallon. The apparent cost disadvantage of fueleconomy standards is even more pronounced in the short run, as fuel taxes give all motoristsan immediate incentive to save fuel by driving less, while new vehicle standards permeate thevehicle fleet gradually.Finally, it appears that market-modeling studies based on historical data may not provide

reliable estimates of the future costs of fuel economy regulations. Cost estimates are sensitiveto assumptions about baseline fuel economy (i.e., without policy changes), which can changesubstantially over time. For example, rising oil prices in the future, or progress on developingfuel-saving technologies, could shift baseline demand toward more efficient vehicles, therebyreducing both the effectiveness and the costs of a given fuel economy standard.

10Anderson and Sallee (2010) do not model consumer and automaker behavior explicitly but instead observethat automakers have often failed to take full advantage of low-cost flexible fuel credits in recent years, whichprovides an indirect estimate of automaker compliance costs.11For example, Kleit (2004) and Austin and Dinan (2005) scale up the marginal cost of adding fuel-savingtechnologies (to proxy for hidden costs) until, in the observed baseline (i.e., prior to the CAFE policy), thereare no fuel-saving technologies that could be profitably adopted.

98 Anderson et al.

at University of C


Dow

nloaded from


Comparing the Welfare Effects of Fuel Economy Standardsand Fuel Taxes

This section presents a stylized model, which accounts for externalities and possible misper-ceptions market failures, to compare the overall welfare effects of fuel taxes and fuel economystandards.

Conceptual Framework

Consider Figure 3, reproduced from Parry, Evans, and Oates (2010), which shows variousmarginal cost curves for reducing long-run gasoline use in the United States below a givenbaseline level through fuel taxes and fuel economy standards. The solid gray curves indicatemarginal costs if there were no market failures or preexisting fuel policies. For the fuel tax, thearea under this curve corresponds to the deadweight loss triangle created by the tax in thegasoline market. With a perfectly elastic fuel supply curve, the slope of the marginal costdepends on the long-run gasoline demand elasticity, taken to be !0.4 (see, e.g., Smalland Van Dender 2007). The marginal cost for the fuel economy standard has a much steeperslope than the curve for the fuel tax. This is because the policy places the entire burden of thefuel reduction on fuel economy improvements and fails to exploit any fuel savings throughreduced vehicle mileage.The dashed gray curves in Figure 3 take into account preexisting fuel taxes, which are

approximately 40 cents per gallon in the United States (with state and federal taxescombined). For both policies, marginal costs now have an intercept equal to this priortax, which reflects the initial wedge between demand and supply prices in the fuel market.That is, with nomarket failures, the preexisting tax would cause motorists to overinvest in fueleconomy and drive too little.Next, we subtract from these costs estimates of fuel- and mileage-related externality ben-

efits. Although highly contentious, a plausible value for global warming damages is $20 perton of CO2, which amounts to 18 cents per gallon of gasoline (see, e.g., Newbold et al. 2009;Tol 2009; Aldy et al. 2010; IWG 2010). Netting out this benefit, the marginal cost curves inFigure 3 shift down, but they would still have positive intercepts of 22 cents per gallon. Thus,under this assumption, global warming by itself is not sufficient to warrant raising the existingfuel tax or imposing fuel economy standards on top of the fuel tax.As for mileage-related externalities, Parry, Evans, and Oates (2010) assume that the mar-

ginal costs of traffic congestion (averaged across region and time of day) are 4.5 cents per mile,the external costs of traffic accidents are 3.5 cents per mile, and local pollution damages are1.0 cents per mile.12 Given current on-road fuel economy levels in the United States (about 22miles per gallon), these combined externalities would be equivalent to about $2 per gallon.However, because only about half of the fuel reduction under fuel taxes comes from reducedmileage (the remainder comes from fuel economy improvements that do not affect theseexternalities), the externality benefit per gallon of fuel savings amounts to about $1 per gallon.

12Roughly speaking, local emissions vary with vehicle miles driven rather than total fuel use. That is, they areindependent of vehicle fuel economy. This is because all new vehicles have to satisfy the same grams-per-milestandards, regardless of fuel economy, and these standards are approximately maintained throughouta vehicle’s life through emissions inspection and maintenance programs.


at University of C


Dow

nloaded from


Overall, the marginal cost curve for the fuel tax (the solid black curve) now has anintercept of about !80 cents per gallon and marginal costs are negative up to a fuelreduction of about 11 percent. This corresponds to the reduction implied by raisingthe fuel tax from its current level to its externality-correcting level (about $1.26 pergallon). Under the fuel economy standard, however, mileage-related externalities increase(albeit moderately) due to the rebound effect. The (solid black) marginal cost for thispolicy now has an intercept of 43 cents per gallon, implying that fuel economy standardsare still welfare reducing.However, there is an additional source of welfare gain if consumers undervalue fuel

economy improvements. Parry, Evans, and Oates (2010) consider a plausible upperbound for the potential magnitude of this misperceptions market failure where abouttwo-thirds of the lifetime fuel savings are not internalized (this corresponds to a casein which consumers consider fuel savings only over the first three years of the life ofa new vehicle). Under the fuel tax, the marginal cost curve (now indicated by the dashedblack line in Figure 3) is shifted down further to the extent it increases fuel economy andaddresses the market failure. However, the downward shift in the marginal cost for thefuel economy standard is even greater since all (rather than part) of the fuel reductionunder this policy comes from improved fuel economy. Nonetheless, the marginal cost forthis policy does not fall below that for the fuel tax.This last finding implies that fuel taxes and fuel economy standards are not complementary

instruments, despite the large misperceptions market failure. That is, reducing gasoline usethrough fuel taxes is more efficient than reducing it through some combination of fuel taxesand fuel economy standards. For any given fuel reduction, tightening the standard (andreducing the tax) implies that more of the reduction will come from better fuel economyand less from reductions in vehicle miles traveled. In Parry, Evans, and Oates (2010), any

Figure 3 Marginal costs of a fuel tax versus a fuel economy standardSource: Parry, Evans, and Oates (2010).

100 Anderson et al.

at University of C


Dow

nloaded from


efficiency benefits from improving fuel economy are offset by efficiency losses, as less is doneto reduce traffic congestion, accidents, and local pollution.13

Some Caveats

One caveat here is that fuel taxes are an inefficient way to reduce the mileage-related exter-nalities. For example, traffic congestion is highly specific to region and time of day and ismuch better addressed through peak-period pricing. If mileage-related externalities werecomprehensively internalized through other policies, it would then be optimal to addressany remaining market failures associated with fuel economy misperceptions through stand-ards rather than through fuel taxes.Furthermore, standards may bemore politically practical than taxes. Motorists appear to be

less opposed to standards because they do not transfer a large amount of revenue to thegovernment. Fuel taxes are also thought to be regressive, at least prior to the use of revenues(West 2004;West andWilliams 2004; Bento et al. 2009). In contrast, standardsmay actually beprogressive, as their direct impact is on new vehicles, which are disproportionately consumedby higher-income families (though, as discussed in Jacobsen 2010b, distributional effects arecomplicated because of secondary effects on used-car prices).14

Suppose fuel economy standards are the only practical option—do they domore good thanharm? Unfortunately, it is difficult to answer this question definitively. If there are largeclimate and energy security benefits from cutting fuel use, and a significant market failureassociated with fuel economy decisions, then standards can be welfare improving. On theother hand, if any benefits from improved energy security, and addressing fuel economymarket failures, are modest, the costs of standards can easily outweigh the benefits.15

Finally, the standards themselves may help create a stable environment for the devel-opment and adoption of fuel-saving technologies with high up front costs and long-termpayoffs. It is important to note that the welfare gains from such induced innovation arenot included in Figure 3. While a higher fuel tax would also provide incentives for innovation,the standard could provide more direct incentives by eliminating the downside risks to

13Accounting for fiscal considerations may further strengthen the welfare basis for fuel taxes over standards.Gasoline appears to be a relatively weak substitute for leisure, which implies that (up to a point) replacinglabor taxes (e.g., income and payroll taxes) with gasoline taxes will increase labor supply and generateadditional welfare gains that are not possible under standards (West and Williams 2005; Parry 2007).14Another potential rationale for using standards rather than taxes is related to the imperfectly competitivenature of vehicle markets, in which a few large manufacturers manage sales of large fleets of vehicles. If fueleconomy can be used as a means to segment consumers, manufacturers may engage in price discriminationin order to garner higher vehicle prices by underproviding fuel economy to consumer types that value it less(e.g., large car buyers) and overproviding it to those that value it more (Fischer 2005, 2010). However, whilethis kind of market failure can potentially be a motivation for using minimum standards as a means oflimiting the availability of price discrimination mechanisms, average fuel economy standards like CAFEdo little to address problems of excessive differentiation.15For example, without a consumer misperceptions market failure, and the above CO2 externality benefit(i.e., $20 per ton), energy security benefits would need to be well above 22 cents per gallon for standards to bewelfare enhancing. One security benefit is the reduced vulnerability of the macroeconomy to oil price shocks.However, Brown and Huntington (2010) put the benefit at only about 5 cents per gallon. More generally,there might be important foreign policy or national security benefits from reducing dependence on importedoil from the Middle East, though these sorts of benefits are extremely difficult to quantify.


at University of C


Dow

nloaded from


innovators from fuel price volatility and more precisely targeting domestic and internationalspillovers associated with fuel-saving technologies (e.g., Barla and Proost 2010). In fact, froma policymaker’s perspective, inducing innovation over the long term may actually be one ofthe most important objectives of regulation.

Standards versus Feebates

There has been growing interest in feebates. Such policies combine a fee for new vehicles thathave fuel economy below some specified ‘‘pivot point’’ with rebates for vehicles that have fueleconomy above the pivot point, where these fees/rebates could be levied at either the con-sumer or the manufacturer level (e.g., Greene et al. 2005; Fischer 2008). Policymakers in theUnited States have been discussing feebates as an alternative to CAFE standards since the early1990s, and feebates have already been implemented (in a modest form) in Ontario in 1991and, more recently, on a national basis in Canada and France. Like fuel economy standards,feebates need not impose a politically unpopular tax burden on motorists. This is becausethey can be made revenue neutral by setting the pivot point in one year slightly above theaverage fuel economy for new vehicles in the previous year.To provide a constant incentive rate for each gallon of fuel saved, the fee or rebate under

a feebate system should be proportional to fuel consumption per mile (rather than fuel econ-omy, in mpg) regardless of whether those improvements are in small or large vehicles. Be-cause an increase of 1 mpg starting at a lower fuel economy rating has a larger impact ongasoline consumption than a 1 mpg increase starting at a higher mpg, a payment schedulebased on mpg would give a disproportionately small rebate (or subsidy) for fuel savings inlow-mpg vehicles (where the potential for fuel economy improvements is greatest).This section compares the cost-effectiveness of fuel economy standards and feebates and

discusses their compatibility with other policy instruments.

Comparison of Cost-Effectiveness

Within a given year, to achieve an average fuel economy target for new vehicles at the lowestindustry-wide cost requires equating marginal compliance costs across automakers. Figure 4shows marginal compliance costs (net of consumer willingness to pay for fuel economyimprovements) for reducing fuel consumption per mile for representative high-cost andlow-cost firms—for example, firms that specialize in large cars and small cars. The indus-try-wide costs of meeting an average standard of !f gallons per mile are minimized when thehigh-cost and low-cost firms reduce fuel consumption per mile to fH and fL, respectively(assuming firms have the same fleet size).Under a traditional fuel economy (or vehicle CO2) standard, all firms are required to meet

the same industry-wide standard. This results in different marginal compliance costs acrossfirms. As shown in Figure 4, this causes an efficiency loss, which is indicated by the differencebetween the taller and shorter shaded trapezoids. Feebates automatically achieve the cost-minimizing outcome, in which high-cost firms pay a fee of s on each unit between theiractual fuel consumption per mile and the target level, while low-cost firms receive a subsidyof s for each unit that their actual fuel consumption per mile exceeds the target level. Austinand Dinan (2005), for example, estimate that the total costs of complying with fuel economy

102 Anderson et al.

at University of C


Dow

nloaded from


targets would fall by a significant (though not dramatic) 15 percent with this equalization ofmarginal compliance costs across firms.The least-cost outcome could also be achieved using fuel economy standards by allowing

limitless trading of credits among firms. In principle, feebates and standards with credit tradingare equivalent instruments. The (uniform) price on fuel economy credits ensures that allvehicles face the same marginal incentive to improve fuel economy, thus playing the same roleas the fee and rebate. Under the regulatory approach, the industry-wide standard plays thesame role as the pivot-point fuel economy in the feebate system. In fact, in the United Statesand elsewhere, inter-firm trading provisions are being implemented, which undermines oneof the key arguments in favor of feebates. However, because there is a relatively small numberof firms in the automobile market, the trading market could be thin, which raises the riskthat limited arbitrage would not be sufficient to equalize marginal compliance costs.Another way to improve the cost-effectiveness of standards in the absence of credit trading is

to set standards based on vehicle size or other attributes, which introduces variation in fueleconomy requirements across firms (e.g., firms specializing in relatively large vehicles receivea less stringent fleet-wide target). However, in the absence of credit trading, the full equalizationof marginal abatement costs will not be realized. Moreover, size-based standards blunt incen-tives for reducing vehicle size, which eliminates a potentially important option for improvingfleet fuel economy.16 Elmer and Fischer (2010) discuss how this problem could be avoided byallowing for automaker-specific targets based on historical (rather than current) fleet attributes.In a multi-period context, cost minimization also requires equating the marginal costs of

compliance in one year with the (discounted) marginal costs of compliance in another.

Figure 4 Potential savings in compliance costs from feebates and flexible regulationsNotes: This figure indicates cost savings from equating marginal compliance costs across firms rather thanforcing firms to meet the same fuel per-mile standard. Cost savings are indicated by the shaded trapezoidunder the higher marginal cost curve less the shaded trapezoid under the lower marginal cost curve.

16The potential importance of downsizing as a means of improving fuel economy is clear if one compares theUnited States with Europe, which enjoys a much higher fleet average due in large part to the predominance ofsmaller vehicles.


at University of C


Dow

nloaded from


From a broader welfare perspective, this requirement presumes that the marginal externalbenefits of fuel reductions are the same over time, which is a fairly reasonable assumption,at least for CO2. In their traditional form, fuel economy standards violate this condition, asthe marginal costs of complying with a fixed fuel economy standard will vary from year toyear with volatility in fuel prices, which affect consumer willingness to pay for moreefficient vehicles, and other factors. There is little evidence, however, on how this lackof dynamic flexibility might increase costs. Furthermore, recent provisions in the CAFEprogram are helping to deal with cost uncertainty, at least in part. Automakers arenow allowed to bank CAFE credits for up to five years when they exceed the fuel economystandard in a given year, and they may borrow credits if they fall short of the standard, solong as credits are paid back within three years. Perhaps more important is the fact thattechnological innovation brings down the costs of complying with a fixed fuel economytarget, which means that marginal costs will not be equated over time unless the standardcontinually increases at an appropriate rate.Feebate programs can easily accommodate both cost uncertainty and technical change. In

periods of high compliance costs, automakers can choose to pay more in fees, or forgosubsidies, enabling them to sell a greater share of vehicles with low fuel economy, and viceversa in periods of low compliance costs. Feebates also provide ongoing incentives forimprovement, as the value of fuel economy stays constant and does not diminish withtechnical advancements.In practice, feebate systems have featured multiple pivot points as a way of lowering the tax

burden paid, or subsidies received, by individual manufacturers. This system results in ‘‘taxnotches,’’ where a marginal change in fuel economy can create a large discrete change in taxtreatment. Sallee and Slemrod (2010) find evidence that manufacturers respond to theseincentives by slightly modifying vehicles close to cutoff points in the tax system, resultingin some loss of efficiency compared to both a smooth schedule and fuel economy standards.Finally, while fuel economy standards are applied only to automakers, feebates could be

applied to automakers, dealers, or consumers. Economic theory suggests that with efficientmarkets, the point of compliance should not matter, as the incentive will be passed alongthrough the price of the vehicle. Evidence in Busse, Silva-Risso, and Zettelmeyer (2006), how-ever, suggests that the point of compliance may be relevant. They find that because ofinformation asymmetries, the behavioral response to price incentives (such as feebates)may be stronger if the incentives are levied at the consumer rather than the producer level,which has potential implications for cost-effectiveness. Politicians appear to believe in thisasymmetry, as fuel economy taxes have historically been levied onmanufacturers while rebateshave been given directly to consumers (Sallee 2010). This is an issue that requires further study.

Compatibility with Other Policy Instruments

The gains from transitioning from standards to feebates may not appear to be very significanton purely cost-effectiveness grounds. However, recent reforms to fuel economy programshave highlighted an important drawback of standards—that they can undermine the perfor-mance of other, increasingly common, policy interventions in the transportation sector.For example, the United States provides generous tax credits for the purchase of hybrid

vehicles. One of the primary objectives of these subsidies is to reduce CO2 emissions and

104 Anderson et al.

at University of C


Dow

nloaded from


dependence on oil. In the presence of binding fuel economy regulations, however, greaterpenetration of hybrid vehicles will simply allow automakers to cut back on fuel-savingtechnologies for conventional gasoline vehicles (McConnell and Turrentine 2010). Sim-ilarly, taxes on vehicles with low fuel economy or high CO2 emissions may increasedemand for smaller, more efficient vehicles, but this will allow automakers to install fewerfuel-saving technologies than would otherwise be needed to meet the standard. Totake another example, under a binding nationwide fuel economy program, a state orregional program that increases fuel economy in one area may be offset by reductionsin fuel economy in other regions (Goulder, Jacobsen, and van Benthem 2009). In fact,this concern recently motivated the federal government in the United States to set stand-ards that are equivalent to the more aggressive standards already being phased in underCalifornia law.In contrast to fuel economy standards, pricing instruments tend to be additive. That is,

hybrid vehicle subsidies and gas guzzler taxes will improve fuel economy and reduce gas-oline use, regardless of any preexisting fuel taxes or the presence of feebates. Similarly,feebates at the national level would not undermine the effects of regional environmentalor fuel economy initiatives.

Conclusions

The future effectiveness of fuel economy standards is difficult to gauge, given that the baselinefuel economy (in the absence of policy) is sensitive to oil prices, technology costs, and otherfactors. In addition, the cost-effectiveness of standards remains contentious, due in particularto uncertainty about how consumers value fuel-saving benefits. At first glance, fuel economystandards appear to be difficult to justify on welfare grounds, given that fuel taxes—even inthe United States—exceedmost estimates for per-gallon climate damages. In fact, our stylizedmodel suggests that high levels of fuel taxation can be defended—up to a point—on eco-nomic efficiency grounds since they reduce congestion and other externalities that vary withmiles driven and are relatively large in magnitude. Even if there is a large market failureassociated with consumer misperceptions about the benefits of fuel economy, and even ifthe social costs of global warming or oil dependence are high, this need not imply a rolefor fuel economy regulations if fuel taxes can be adjusted.On the other hand, in countries such as the United States, standards appear to be more

acceptable to the public, and hence more practicable, than high fuel taxes. Standards may alsohelp create a more stable environment for the development of clean technology by removingsome of the downside risks to innovators in a world of uncertain fuel prices. However, it is notwell understood whether fuel economy standards are better or worse than other instruments,such as technology prizes, fuel taxes, and fuel price floors, in encouraging such innovation.While the appropriate role of fuel economy standards remains an unresolved issue in the eco-

nomics literature, recent structural reforms of existing programs, particularly provisions thatexpand opportunities for credit trading across firms, vehicle types, and over time, have helpedto improve the cost-effectiveness of standards. However, their relative incompatibility with otherinterventions in the vehicle market and at different levels of government remains a concern. Thisis one important reason why feebate systems deserve further consideration and research.


at University of C


Dow

nloaded from


References

Aldy, Joseph, Alan J. Krupnick, Richard G. Newell,Ian W. H. Parry, and William A. Pizer. 2010.Designing climate mitigation policy. Journal ofEconomic Literature 48: 903–34.

Allcott, Hunt, and Nathan Wozny. 2009. Gasolineprices, fuel economy, and the energy paradox.Discussion Paper, Department of Economics,Massachusetts Institute of Technology.

An, Feng, and Amanda Sauer. 2004. Comparison ofpassenger vehicle fuel economy and greenhouse gasemission standards around the world. Arlington, VA:Pew Center on Global Climate Change.

Anas, Alex, and Robin Lindsey. 2011. Reducingurban road transportation externalities: Roadpricing in theory and in practice. Review ofEnvironmental Economics and Policy 10.1093/reep/req019.

Anderson, Soren, and James Sallee. Forthcoming.Using loopholes to reveal the marginal cost ofregulation: The case of fuel-economy standards.American Economic Review.

Austin, David, and Terry Dinan. 2005. Clearing theair: The costs and consequences of higher Istandards and increased gasoline taxes. Journal ofEnvironmental Economics and Management 50:562–82.

Barla, Philippe, and Stef Proost. 2010. Automobilefuel efficiency policies with international costspillovers. Working Paper, Center for EconomicStudies, KU Leuven, Belgium.

Bento, Antonio, Larry Goulder, Mark Jacobsen, andRoger von Haefen. 2009. Distributional andefficiency impacts of increased U.S. gasoline taxes.American Economic Review 99: 667–99.

Brown, Stephen P. A., and Hillard G. Huntington.2010. Reassessing the oil security premium.Discussion Paper 10-05, Resources for the Future,Washington, DC.

Busse, Meghan, Jorge Silva-Risso, and FlorianZettelmeyer. 2006. $1,000 cash back: The pass-through of auto promotions. American EconomicReview 96: 1253–70.

Creyts, Jon, Anton Derkach, Scott Nyquist, KenOstrowski, and Jack Stephenson. 2007. ReducingU.S. greenhouse gas emissions: How much at whatcost? Washington, DC: McKinsey & Company.

Davis, Stacy. 1999. Transportation energy fact book.Edition 19. Oak Ridge, TN: Center for

Transportation Analysis, Oak Ridge NationalLaboratory.

Dreyfus, Mark K., and W. Kip Viscusi. 1995. Ratesof time preference and consumer valuations ofautomobile safety and fuel efficiency. Journal of Lawand Economics 38: 79–105.

Dubin, J. A., and D. I. McFadden. 1984. Aneconometric analysis of residential electricappliance holdings and consumption. Econometrica52: 345–62.

EIA. 2009.Annual energy outlook 2009. Washington,DC: Energy Information Administration, U.S.Department of Energy.

Elmer, Carl-Friedrich, and Carolyn Fischer. 2010.Designing vehicle CO2 emission standards:Uniform vs. weight-based standards. WorkingPaper, Resources for the Future, Washington, DC.

EPA. 2009. Light-duty automotive technology, carbondioxide emissions, and fuel economy trends: 1975through 2009. EPA-420-R-09-014. Washington, DC:U.S. Environmental Protection Agency.

Fischer, Carolyn. 2005. On the importance of thesupply side in demand-side management. EnergyEconomics 27 (1): 165–80.

———. 2008. Comparing flexibility mechanismsfor fuel economy standards. Energy Policy 36:3116–24.

———. 2010. Imperfect competition, consumerbehavior, and the provision of fuel efficiency inlight-duty vehicles. Working Paper, Resources forthe Future, Washington, DC.

Fischer, Carolyn, Winston Harrington, and Ian W.H. Parry. 2007. Should corporate average fueleconomy standards be tightened? Energy Journal 28:1–29.

Gayer, Ted. 2004. The fatality risks of sport-utilityvehicles, vans, and pickups relative to cars. Journalof Risk and Uncertainty 28: 103–33.

Goldberg, Pinelopi Koujianou. 1998. The effects ofthe corporate average fuel efficiency standards inthe US. Journal of Industrial Economics 46: 1–33.

Goulder, Lawrence H., Mark R. Jacobsen, andArthut A. van Benthem. 2009. Unintendedconsequences from nested state & federalregulations: The case of the Pavley greenhouse-gas-per-mile limits. Working Paper No. 15337, NationalBureau of Economic Research, Cambridge, MA.

106 Anderson et al.

at University of C


Dow

nloaded from


Gramlich, Jacob. 2008. Gas prices and endogenousproduct selection in the U.S. automobile industry.Working Paper, Department of Economics, YaleUniversity.

Greene, David L. 2010. How consumers value fueleconomy: A literature review. EPA-420-R-10-008.Washington, DC: U.S. Environmental ProtectionAgency.

Greene, David L., Philip D. Patterson, MargaretSingh, and Jia Li. 2005. Feebates, rebates, and gas-guzzler taxes: A study of incentives for increasedfuel economy. Energy Policy 33: 757–75.

Hausman, Jerry A. 1979. Individual discount ratesand the purchase and utilization of energy-usingdurables. Bell Journal of Economics 10: 33–54.

Helfand, Gloria, and Anne Wolverton. 2009.Evaluating the consumer response to fuel economy:A review of the literature. Discussion Paper,National Center for Environmental Economics,U.S. Environmental Protection Agency.

ICCT. 2009. Passenger vehicle greenhouse gas and fueleconomy standards: A global update. Washington,DC: International Council on CleanTransportation.

IEA. 2009. CO2 emissions from fuel consumption.Paris France: International Energy Agency.

———. 2010. Review of international policies forvehicle fuel efficiency. Paris France: InternationalEnergy Agency.

IWG. 2010. Social cost of carbon for regulatoryimpact analysis under executive order 12866.Washington, DC: Interagency Working Group onSocial Cost of Carbon.

Jacobsen, Mark. 2010a. Evaluating U.S. fueleconomy standards in a model with producer andhousehold heterogeneity. Discussion Paper,University of California, San Diego.

———. 2010b. Fuel economy, car class mix, andsafety. Discussion Paper, University of California,San Diego.

Kahn, and James A. 1986. Gasoline prices and theused automobile market: A rational expectationsasset price approach.Quarterly Journal of Economics101: 323–40.

Kilian, Lutz, and Eric R. Sims. 2006. The effects ofreal gasoline prices on automobile demand: Astructural analysis using micro data. Unpublishedmanuscript, University of Michigan.

Klier, Thomas, and Joshua Linn. 2010. New vehiclecharacteristics and the cost of the corporate averagefuel economy standard. Discussion Paper,Resources for the Future, Washington, DC.

Kleit, Andrew N. 2004. Impacts of long-rangeincreases in the corporate average fuel economy(CAFE) standard. Economic Inquiry 42: 279–94.

Knittel, Christopher R. 2009. Automobiles onsteroids: Product attribute trade-offs andtechnological progress in the automobile sector.NBER Working Paper No. 15162.

Larrick, Richard P., and Jack B. Soll. 2008. TheMPG Illusion. Science 320: 1593–4.

Li, Shanjun. Forthcoming. Traffic safety and vehiclechoice: Quantifying the effects of the ‘‘arms race’’ onAmerican roads. Journal of Applied Econometrics.

McConnell, Virginia D., and Tom Turrentine. 2010.Technical report: Hybrid vehicles and policies toreduce GHG emissions. Working Paper, Resourcesfor the Future, Washington, DC.

Newbold, StephenC.,CharlesGriffiths,ChrisMoore,andAnnWolverton. 2009. The ‘‘social cost of carbon’’made simple. Discussion Paper, National Center forEnvironmental Economics, U.S. EnvironmentalProtection Agency, Washington, DC.

NHTSA. 2009. Summary of fuel economyperformance. Washington, DC: National HighwayTraffic Safety Administration, U.S. Department ofTransportation.

———. 2010. Final regulatory impact analysis,corporate average fuel economy for MY 2012–MY2016 passenger cars and light trucks. Washington,DC: National Highway Traffic SafetyAdministration, U.S. Department ofTransportation.

NRC. 2002. Effectiveness and impact of corporateaverage fuel economy (CAFE) standards. NationalResearch Council. Washington, DC: NationalAcademy Press.

Parry, Ian W. H. 2007. Are the costs of reducinggreenhouse gases from passenger vehicles negative?Journal of Urban Economics 62: 273–93

Parry, Ian W. H, David Evans, andWallace E. Oates.2010. Are energy efficiency standards justified?Discussion Paper, Resources for the Future,Washington, DC.

Proost, Stef, and Kurt Van Dender. 2011. Whatlong-term road transport future? Trends and policyoptions. Review of Environmental Economics andPolicy 10.1093/reep/req022.


at University of C


Dow

nloaded from


Sallee, James M. Forthcoming. Fuel economytaxation. Tax Policy and the Economy 25.

Sallee, James M., and Joel Slemrod. 2010. Carnotches. Working Paper, University of Chicago.

Sallee, James M., Sarah E. West, and Wei Fan. 2009.Consumer valuation of fuel economy: A microdataapproach. Proceedings of the National TaxAssociation Annual Conference on Taxation. 250–55.

Small, Kenneth A. 2010. Energy policies forpassenger transportation: A comparison of costsand effectiveness. Discussion Paper, University ofCalifornia, Irvine.

Small, Kenneth A., and Kurt Van Dender. 2007. Fuelefficiency and motor vehicle travel: The decliningrebound effect. Energy Journal 28: 25–52.

Tol,RichardS. J. 2009.Theeconomiceffectsofclimatechange. Journal of Economic Perspectives 23: 29–51.

Turrentine, Thomas S., and Kenneth S. Kurani.2007. Car buyers and fuel economy. Energy Policy35: 1213–23.

West, Sarah E. 2004. Distributional effects ofalternative vehicle pollution control policies.Journal of Public Economics 88: 735–57.

West, Sarah E., and Roberton C. Williams. 2004.Estimates from a consumer demand system:Implications for the incidence of environmentaltaxes. Journal of Environmental Economics andManagement 47: 535–58.

———. 2005. The cost of reducing gasolineconsumption. American Economic Review Papersand Proceedings 95: 294–99.

White, Michelle. 2004. The ‘‘arms race’’ onAmerican roads: The effect of sport utility vehiclesand pickup trucks on traffic safety. Journal of Lawand Economics 47 (2): 333–56.

Whitefoot, Katie, Meredith Fowlie, and StevenSkerlos. 2010. Evaluating U.S. reformed corporateaverage fuel economy standards using anengineering attribute selection model. WorkingPaper, University of Michigan.

108 Anderson et al.

at University of C


Dow

nloaded from


automobile fuel economy standards: impacts, efﬁciency, and ...nature.berkeley.edu › ~sallee ›...

Documents